Launched my first product today!
AI labs drop a new model almost every 2 weeks.
To switch between them (or between sub-agents) without re-explaining everything, you need memory that travels with you.
Meet Onadit: local memory on your Mac.
Conversations get saved, split and indexed (keywords + embeddings). Your agents (Claude Code, Codex, Cursor… via MCP) can pull from it.
Ask “what stack did I choose for the content pipeline, no infra?”
→ finds: “No database. No server. No build step. Just markdown, JSON, and Claude”
→ from ~63k tokens to 820 (imagine the savings + efficiency when agents work in chain)
https://t.co/CtLlErWceB (first 7 days free)
First launch so would mean a lot if you try it or share it 🙏
@saikhamhein the user-words-only evidence rule is smart. do sessions stay searchable after a dream? most of my recalls are "what did we decide on X", not rules so I'm genuinely asking (disclaimer: i've built myself for 9 months now a local agent memory and i'm looking to improve it)
@Augustbuilds I talk to 1 "agent lead" per repo, it spawns the sub agents (Opus, Fable, Sol, Astra, whatever) and since it's always idle i just keep adding work as it comes!
Rex is damn good. On the mac, on the phone, over tailscale, it just works so nice (although in Beta)!
Yes it's a terminal so mono font hater will hate, but for the one who were already ok with Ghossty, Kitty or any other, it's a blast!
@Kemonocept@raroque same skill here, claude and codex call each other headless. the gotcha was codex's sandbox blocking claude's network and ~/.claude login :)